An Injury-like Signature of the Extracellular Glioma Metabolome

Abstract

Aberrant metabolism is a hallmark of malignancies including gliomas. Intracranial microdialysis enables the longitudinal collection of extracellular metabolites within CNS tissues including gliomas and can be leveraged to evaluate changes in the CNS microenvironment over a period of days. However, delayed metabolic impacts of CNS injury from catheter placement could represent an important covariate for interpreting the pharmacodynamic impacts of candidate therapies. Intracranial microdialysis was performed in patient-derived glioma xenografts of glioma before and 72 h after systemic treatment with either temozolomide (TMZ) or a vehicle. Microdialysate from GBM164, an IDH-mutant glioma patient-derived xenograft, revealed a distinct metabolic signature relative to the brain that recapitulated the metabolic features observed in human glioma microdialysate. Unexpectedly, catheter insertion into the brains of non-tumor-bearing animals triggered metabolic changes that were significantly enriched for the extracellular metabolome of glioma itself. TMZ administration attenuated this resemblance. The human glioma microdialysate was significantly enriched for both the PDX versus brain signature in mice and the induced metabolome of catheter placement within the murine control brain. These data illustrate the feasibility of microdialysis to identify and monitor the extracellular metabolome of diseased versus relatively normal brains while highlighting the similarity between the extracellular metabolome of human gliomas and that of CNS injury.

Keywords: CNS injury; glioma; microdialysis.

Figure 1

Schematic of experimental design. Following TMZ or vehicle (1X PBS) administration, microdialysate was collected at two different time points, once at baseline and after 72 h. With two variables of tissue type (tumor and brain) and treatment option (TMZ or vehicle), there were four experimental conditions. This figure was created with BioRender.com.

Figure 2

Microdialysis of IDH-mutant high-grade glioma versus normal brain. (A) Heat map of top and bottom 50 differentially abundant raw metabolite peak values for both brain and tumor based on fold changes between the means. (B) Volcano plot of fold changes and Mann–Whitney U test of fold changes calculated between IDH-mutant PDX and brain at baseline (n = 9 for tumor; n = 12 for brain) were used to construct a tumor versus brain volcano plot (cut-offs for significance marked as dashed lines: p-value ≤ 0.05; FC ≥ 2). (C) Fold changes and p-values (based on Mann–Whitney U test) of 3 most significantly altered metabolites.

Figure 3

Metabolic signatures of vehicle and TMZ in brain and tumor (72 h vs. baseline) (A) For each mouse with a vehicle-treated brain, the fold changes of the peak area were calculated for each metabolite (72 h vs. baseline) and ranked from highest to lowest. These ranked lists were then ordered based on the average of the experimental cohort (n = 3). Inset: NES values from enrichment analyses performed between the total ranked metabolite list of each individual mouse and a library composed of the top and bottom 35 metabolite lists of each mouse. (B) The same analyses performed in A were applied to mice with vehicle-treated tumors. (C) For each mouse with a TMZ-treated brain, the fold changes of the peak area were calculated for each metabolite (72 h vs. baseline) and ranked from highest to lowest. These ranked lists were then ordered based on the average of the experimental cohort (n = 3). Inset: NES values from enrichment analyses performed between the total ranked metabolite list of each individual mouse and a library composed of the top and bottom 35 metabolite lists of each mouse. (D) The same analyses performed in C were applied to mice with TMZ-treated tumors.

Figure 4

Defining the extracellular metabolome of tumor (A) Ranked lists of the GBM164 vs. brain signature at baseline, 72 h after vehicle administration, and 72 h after TMZ treatment were created by evaluating the average fold changes in metabolites between brain and tumor at the respective time points and then ranking metabolites from highest to lowest fold changes. These ranked lists were ordered based on the GBM164 vs. brain signature at baseline. (B) Plots of enrichment analysis between the baseline GBM164 vs. brain ranked list and metabolite sets of (i) GBM164 vs. brain after 72 h of vehicle administration and (ii) GBM164 vs. brain after 72 h of TMZ treatment. (C) Plot of enrichment analysis between the ranked list of GBM164 vs. brain after 72 h of vehicle administration and the top 35 metabolites of GBM164 vs. brain after 72 h of TMZ treatment. (D) NES values from enrichment analyses performed between each ranked list and a library composed of the top and bottom 35 metabolites of every list.

Figure 5

Enrichment with intra-operatively acquired human glioma microdialysate (A) Enrichment analysis was utilized to determine the enrichment of eight human patients’ enhancing tumor vs. brain ranked lists for the metabolite sets of catheter-induced injury, PDX vs. brain, and catheter-induced injury in the setting of TMZ. Positive normalized enrichment scores (NESs) indicate metabolic similarities to these signatures. ANOVA and pairwise comparisons of normalized enrichment scores calculated between the average patient glioma signature and the ranked lists of catheter-induced injury (72 h vs. baseline in vehicle-treated brain), the glioma PDX signature at baseline, and the injury signature after TMZ treatment (72 h vs. baseline in TMZ-treated brain) are shown. All NES values above the dashed line had FDR ≤ 0.05. (B) Enrichment plots of average patients’ enhancing tumor vs. brain ranked list and the top 35 metabolites of the injury signature, glioma PDX signature at baseline, and the injury signature after TMZ. (C) Fold changes of metabolites at leading edge of enrichment in injury signature and average patient tumor vs. brain signature.